1. Volume 26, Issue 3, Pages 446-458.e8 (March 2018)
Disease Variants of FGFR3 Reveal Molecular Basis for the Recognition and Additional Roles for Cdc37 in Hsp90 Chaperone System 
Tom D. Bunney, Alison J. Inglis, Domenico Sanfelice, Brendan Farrell, Christopher J. Kerr, Gary S. Thompson, Glenn R. Masson, Nethaji Thiyagarajan, Dmitri I. Svergun, Roger L. Williams, Alexander L. Breeze, Matilda Katan 
Structure 
Volume 26, Issue 3, Pages e8 (March 2018)
DOI: /j.str
Copyright © 2018 The Authors Terms and Conditions

2. Structure 2018 26, 446-458.e8DOI: (10.1016/j.str.2018.01.016)
Copyright © 2018 The Authors Terms and Conditions

3. Figure 1
Interaction of FGFR Kinase Domains and FGFR3 Variants with Cdc37 and Hsp90
(A) SEC of isolated components Hsp90, Cdc37, and FGFR3WT (top panels) and of their mixture (ratio 2:1:1), showing the ternary complex (fractions T1-T3) and free FGFR3WT (fractions F1-F3) (bottom panel). The graphs (below) show quantitation of the incorporation of kinases in the ternary complex for FGFR1, 2, 3, and 4 (left), and for sB-RafV600E (right). The error bars represent the SD (n = 3).
(B) Comparison of FGFR3WT with sB-RafV600E using a pull-down binding assay using immobilized Cdc37. A quantification of the amount of kinase bound is shown in the lower panel.
(C) SEC of FGFR3E466K (top panel). SEC of a 1:1 mixture with Cdc37 shows the binary complex (fraction B1-B3) and free kinase (fractions F1-F3) (middle panel) and of a mixture with both Hsp90 and Cdc37 (ratio 2:1:1) showing the ternary complex (fraction T1-T3), a small amount of the binary complex (fractions B1-B3) and free kinase (fractions F1-F3) (bottom panel). The graph (below) shows quantification of FGFR3 in the ternary complex for the FGFR3WT, FGFR3E466K, FGFR3I538F, FGFR3N540K, and FGFR3K650E. The error bars represent the SD (n = 3).
(D) Binding of FGFR3E466K to immobilized Cdc37 (left) and comparison of binding of FGFR3WT with FGFR3E466K, FGFR3I538F, FGFR3N540K, and FGFR3K650E in this pull-down assay (right). The error bars represent the SD (n = 3).
(E) Comparison of melting temperatures of kinase domains of FGFR1-4, sB-RafV600E, and FGFR3E466K, FGFR3I538F, FGFR3N540K, and FGFR3K650E. The error bars represent the SD (n = 4).
(F) Recognition of FGFR3WT, FGFR3E466K, FGFR3I538F, FGFR3N540K, and FGFR3K650E by a polyclonal antibody prepared against a peptide corresponding to FGFR3 amino acid residues 467–495 (LS-C214248, Lifespan BioSciences). See also Figures S1–S4.
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions

4. Figure 2
Comparison of Overall Structural Differences between Client Kinase Domains
(A) The FGFR allosteric network running through the N-lobe and linking to the activation loop and the lobe interface. Cartoon representation of an FGFR3WT model, with amino acid residues corresponding to sites of disease-linked mutations found to destabilize the kinase domain colored blue and all other residues within the network shown as pink.
(B) Normalized thermal-shift curves for the indicated kinase domain variants.
(C) A global deuteration profile of FGFR3 proteins. A linear plot of deuteration levels across FGFR3WT (blue) and two variants FGFR3E466K (red) and FGFR3I538F (green). The percentage of deuteration at the 0.3 s time point for each peptide is plotted against the peptide mid-point (i); secondary structure and key features of the kinase domain region are also shown and mutations indicated by dotted lines (bottom). The data plotted are means ± SD (n = 3). The error bars smaller than the icon are omitted.
(D) A global deuteration profile for the kinase domain of sB-RafV600E; the plot (top) and secondary structure (bottom) are represented as in (C).
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions

5. Figure 3 Comparison of FGFR3 Variants by NMR
(A) Superimposed 2D amide TROSY-HSQC experiments for FGFR3WT (blue) and FGFR3I538F mutant (green). Comparison was performed at 283K. Highlighted in four boxes (1–4) are the most prominent differences; new sharp peaks in box 1 are indicated by (∗).
(B) NMR changes were quantified by CSP analysis. The histogram shows in green the successfully assigned residues transferred from WT at 298K.
(C) FGFR3 residues (shown as side chains and labeled) involved in the allosteric network encompassing the N-lobe and the N-lobe/C-lobe interface are highlighted in a cartoon and surface representation of the kinase. The enlarged N-lobe/C-lobe interface is shown in the inset below with F638 (from the DFG motif), I538 (part of the “DFG latch”), and N540 (from the “molecular brake”) highlighted. Underlined residues (E466 within the N-terminal part of the N-lobe and I538 and N540 at the interface), corresponding to mutations that destabilize FGFR3 kinase domain, are shown as a part of an extensive, interconnected allosteric network highlighted by NMR changes. See also Figure S5.
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions

6. Figure 4
Global Changes in Kinase Domains Following Interaction with Cdc37
(A) The difference in HDX for FGFR3 variants in their free form and when in complex with Cdc37. Percent deuteration change was calculated by subtracting the deuteration of each peptide of FGFR3 complexed with Cdc37 from the same peptide of free FGFR3 at the 0.3 s time point. The data plotted are means ± SD (n = 3).
(B) The same depiction as in (A) for sB-RafV600E.
(C) Ribbon representations of FGFR3 and sB-RafV600E kinase domains with HDX differences mapped onto them. Peptides with significant changes (<5%) upon addition of excess Cdc37 have been colored according to the key. For each peptide, the time point with the largest difference between the free and bound states was used in this analysis. See also Figure S6 and Table S1.
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions

7. Figure 5
Changes in FGFR3 Following the Interaction with Cdc37 Assessed by NMR
(A) Selected region of 1H-15N TROSY-HSQC spectra of FGFR3I538F in the absence and presence of 0.5 equivalents of Cdc37, acquired at 283K (top subpanel); for comparison, spectra of FGFR3WT (navy) and FGFR3I538F (lime), both in the presence of 0.7 equivalents of Cdc37 and acquired at 288K are also shown (bottom subpanel). Coloring of the top subpanel is as in panel (A) The location of sharp peaks (1–12, numbered according to relative intensity) is indicated in both panels and red crosses show approximate locations.
(B) Selected spectral regions of FGFR3I538F in the absence (top subpanel) and presence (bottom subpanel) of 0.5 equivalents of Cdc37, illustrating the presence of sharp peaks 1, 2, and 7 in the free state and their increased intensity in the complex.
(C) Peak heights of sharp peaks 1–12 in FGFR3I538F in the absence (green) and presence of 0.25 (maroon) and 0.5 (yellow) equivalents of Cdc37. Assignments of these peaks are given below their respective peak number.
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions

8. Figure 6
Changes in Cdc37 Following FGFR3 Kinase Binding and a Structural Model for the Cdc37/FGFR3 Kinase Domain Binary Complex
(A) The difference in HDX for Cdc37 in its free form and when in complex with FGFR3E466K. For the regions covered (residues 29–119 are lacking in the analysis), percent deuteration change was calculated by subtracting the deuteration of each peptide of Cdc37 complexed with FGFR3 from the same peptide of free Cdc37 at the 0.3 s time point (left panel). Data plotted are the means ± SD (n = 3). Ribbon representations of a Cdc37 model (based on structures and SAXS data-SASDBP9) with HDX differences mapped onto them. Peptides with significant changes (<5%) upon addition of excess FGFR3 have been colored according to the key. For each peptide, the time point with the largest difference between the free and bound states was used in this analysis (right panels). Region of Cdc37 outlined with a dotted line did not yield any data from the HDX-MS analysis.
(B) Surface representations of Cdc37 and FGFR3 kinase domain (depicted at the same scale) with HDX differences between the free and complex forms mapped and colored according to the key in (A).
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions

9. Figure 7
A Structural Model for the Cdc37/FGFR3 Kinase Domain Binary Complex
SAXS envelope for the Cdc37/FGFR3 complex, with Cdc37 in dark and FGFR3 in light gray (top). The best fit for the complex is shown below as cartoon and surface representations of the kinase domain (green) and Cdc37 (purple). Notable features are labeled and regions identified as protected by HDX-MS shaded in corresponding colors. See also Figure S7 and Table S2.
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions

10. Figure 8 Functional Consequences of Cdc37 Binding to FGFR3 Kinases
(A) Western blot utilizing antibodies that recognize phosphorylated tyrosine residues on FGFR3. Increasing ratios of Cdc37 to kinase are monitored for the effect on FGFR3WT, FGFR3E466K, and FGFR3I538F auto-phosphorylation.
(B) As (A) but the effect of increasing urea concentration is monitored for its effect on auto-phosphorylation of FGFR3 variants.
(C) As (A) but the effect of increasing Hsp90/Cdc37 to kinase ratio is monitored for its effect on auto-phosphorylation of FGFR3 variants.
(D) Decrease in normalized kinase auto-phosphorylation (pY654), following addition of Cdc37 (top), urea (middle), or Cdc37 together with Hsp90 (bottom). Data plotted are the means ± SD (n = 3).
(E) Temperature dependence of client kinase binding to Cdc37 and Hsp90. Immobilized Cdc37 or Hsp90 are utilized to pull down FGFR3 variants at various temperatures (4°C, 15°C, 25°C, and 40°C). Bound protein is visualized using western blotting with a FGFR3 antibody.
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions

11. Figure 9
Insights into Formation, Architecture, and Functional Consequences of the Kinase/Cdc37 Complex
Molecular determinants for differences between strong and weak clients arise from the degree of connectivity within the N-lobe allosteric network (top left). Interaction sites, identified on strong client kinases and Cdc37, mediate formation of stable binary complexes with defined structural organization. Interaction surfaces involve two areas, one formed by interaction between the C-terminal part of Cdc37 and the kinase C-lobe, and another between the N terminus of Cdc37 and a lobe interface/N-lobe of a kinase (center, top). Importantly, Cdc37 binding, via allostery, causes extensive remodeling affecting the entire N-lobe and results in kinase inactivation (center, bottom). Because the overall changes are substantial and the less compact, remodeled regions of the kinase left exposed in the binary complex, the main consequence of this re-modelling is to contribute to direct recognition by Hsp90 (bottom left). Within the ternary complex the kinase becomes further unfolded and interactions with, and relative orientation to, Cdc37 differ. Ultimately, the kinase is protected and refolded.
Structure  , e8DOI: ( /j.str )
Copyright © 2018 The Authors Terms and Conditions